摘要
绝大多数基于机器学习的时间序列分析方法都是人工专家提取特征后进行分析,随着深度学习的快速发展,端到端的方式在时间序列分析领域的应用越来越多,同时整体的方法也变得更加成熟。因此,本文对近年来基于深度学习的时间序列分析方法进行讨论,从应用,网络架构以及思想等方面总结了最新的时间序列预测、分类以及异常检测等任务的深度学习方法,为了解时间序列深度学习解决方案的技术以及发展趋势提供了参考。
Most of the time series analysis methods based on machine learning are analyzed by artificial experts.With the rapid development of deep learning,the end-to-end method is applied more and more in the field of time series analysis,and the overall method is also Become more mature.Therefore,this paper discusses the method of time series analysis based on deep learning in recent years,and summarizes the latest deep learning methods of time series prediction,classification and anomaly detection from the aspects of application,network architecture and ideas,in order to understand the depth of time series.The technology and development trends of learning solutions provide a reference.
出处
《信息技术与信息化》
2019年第1期71-76,共6页
Information Technology and Informatization
关键词
时间序列
深度学习
预测
分类
异常检测
Time Series
Deep Learning
Review
Predict
Classfication
Anomaly Detection
